Project

Back to overview

New approaches to cognition: Developing and testing a multifaceted view on similarity

Applicant Seitz Florian
Number 195389
Funding scheme Doc.CH
Research institution Fakultät für Psychologie Universität Basel
Institution of higher education University of Basel - BS
Main discipline Psychology
Start/End 01.09.2020 - 31.08.2024
Approved amount 268'973.00
Show all

Keywords (10)

similarity; categorization; distance; metrics; exemplar models; formal models; cognitive modeling; machine learning; cognitive and decision science; cognitive psychology

Lay Summary (German)

Lead
Ähnlichkeit ist zentral für viele Prozesse des Denkens wie Kategorisierungen und Entscheidungen. Welche kognitiven Mechanismen psychologischer Ähnlichkeit zugrunde liegen, ist jedoch bis heute nicht abschliessend beantwortet und bedarf weiterer Forschung.
Lay summary

Inhalt und Ziel des Forschungsprojekts
Vergangene Studien nahmen traditionellerweise an, dass Menschen die Ähnlichkeit zweier Objekte durch die Unterschiede in ihren Merkmalen einschätzen. Dieser euklidische Ansatz kann psychologische Ähnlichkeit jedoch nur begrenzt erklären, da er einige empirisch nicht bestätigte Vorannahmen trifft. So sind Objektmerkmale gemäss euklidischem Ansatz zwar voneinander unabhängig, korrelieren jedoch in Wirklichkeit, was Menschen auch berücksichtigen können.
Dieses Projekt prüft, ob psychologische Ähnlichkeit weitere kognitive Mechanismen umfasst, und zielt so auf die Erarbeitung eines vielfältigen Ähnlichkeitskonzepts ab. Nach einer theoretischen Abhandlung der Erfolge und Grenzen der euklidischen Ähnlichkeit in der kognitiven Psychologie werden alternative Ähnlichkeitsansätze empirisch getestet. Es wird untersucht, wie situationsspezifisch psychologische Ähnlichkeit ist, ob ein alternatives Ähnlichkeitsmass eine verbreitete Urteilsverzerrung erklären kann und welche kognitiven Mechanismen psychologischer Ähnlichkeit zwischen Objekten von reellen, natürlichen Kategorien zugrunde liegen.

Wissenschaftlicher und gesellschaftlicher Kontext
Dieses Projekt generiert neue Erkenntnisse zu psychologischer Ähnlichkeit und beeinflusst so ähnlichkeitsbasierte Theorien der Psychologie. Darüber hinaus ist das Projekt durch die Vereinigung von kognitiver Modellierung und maschinellem Lernen stark interdisziplinär geprägt und wirkt sich so auf verwandte Arbeiten in den Computerwissenschaften aus.

Direct link to Lay Summary Last update: 13.05.2022

Responsible applicant and co-applicants

Employees

Publications

Publication
People are Insensitive to Within-Category Feature Correlations in Categorization
SeitzFlorian I., JareckiJana B., RieskampJörg (2021), People are Insensitive to Within-Category Feature Correlations in Categorization, in Proceedings of the 19th International Conference on Cognitive Modelling, University Park, PA: Applied Cognitive Science Lab, Penn State.

Collaboration

Group / person Country
Types of collaboration
Professor Bradley Love, University College London Great Britain and Northern Ireland (Europe)
- in-depth/constructive exchanges on approaches, methods or results
- Publication
- Exchange of personnel

Scientific events

Active participation

Title Type of contribution Title of article or contribution Date Place Persons involved
8th PhD Conference (2nd online) PhD Program Social, Economic, and Decision Psychology, University of Basel Talk given at a conference Sensitivity to Feature Variability Within Categories 27.01.2022 Basel, Switzerland Seitz Florian;
19th International Conference on Cognitive Modelling Poster People are Insensitive to Within-Category Feature Correlations in Categorization 08.07.2021 online Veranstaltung, Switzerland Seitz Florian;
7th PhD Conference (1st online) PhD Program Social, Economic, and Decision Psychology, University of Basel Talk given at a conference Correlational Sensitivity in Exemplar-Based Categorization 21.01.2021 Basel, Switzerland Seitz Florian;


Abstract

Current state of research and project motivation. Similarity lies at the heart of many higher-order cognitive processes such as categorization and judgments. To date, however, the mechanisms by which people represent similarity in judgment and decision making are not well understood. Past research has traditionally followed the Euclidean view on similarity, assuming that the similarity of objects is a function of their differences on the features which constitute the objects. However, the Euclidean similarity has several limitations, including that it treats features separately and is insensitive to correlations among features, whereas people can be sensitive to correlated features. Therefore, further research is needed to uncover how people represent similarity. Main research question and aim. I argue for a multifaceted view on similarity, which emphasizes that the psychology of similarity comprises multiple useful cognitive similarity mechanisms beyond the Euclidean similarity. My thesis will develop the multifaceted view in four connected theoretical and empirical lines of work, aiming at understanding the processes of psychological similarity representation in higher-order cognition. Projects. Project 1 reviews the success and limitations of the Euclidean similarity in cognitive psychology and how machine learning can enrich current psychological theories. In the remaining three projects, I empirically test new hypotheses derived from the multifaceted view on similarity. Project 2 tests how people adapt the way they represent similarity to the current situation. Project 3 tests if an alternative similarity representation can explain a prominent judgmental bias, the conjunction fallacy. Project 4 tests how people represent similarity in real image classification by combining machine learning and cognitive modeling on big data.Method, expected results, and impact. I use an open-science, open-data, open-code approach. I combine experimental data with inferential Bayesian statistics and computational cognitive modeling, relying on my modeling expertise from my Master’s thesis. My thesis will shed light on human learning, adaptation, and representation of similarity and impact the theories in cognitive and decision sciences and general psychology that are based on psychological similarity. The Ph.D. will result in four publications in international peer-reviewed outlets in the fields of cognitive and decision sciences, general psychology, and machine learning.
-